[ipfs-geo]

 This repo is a research of IPFS-GEO (a geography relation object model demo with ipfs). Feel free to star and fork.

Any comments, suggestions? Let us know!. We love PRs :) Please take a look at the Contributing guidelines before opening one.

Explanation

English | 中文

0. Summarize

0.0 Project Goal

This project goal is to build a relation pattern with position information and block chain. Implement a ecological model for people->position->real world->transfer trust->transfer worth->position->people and use block chain to index real world.

0.1 Noun Explaination

• DDApp(Data Decentered Application): For resolve DApp can't rely on the center API, DDApp is a new concept that only the data is decentration. This concept not only can communicate with each other, but also can deploy independent and support p2p transfer.

• IPFS(InterPlanetary File System): An distributed file system.

• GeoHash: The algorithm can greater increase the searching efficiency in large position data, and support convenient cache for app.

1. System Design

1.0 Architecture Design

1.1 Object Model Disign

property	type	comment
geo_id	INT	uniqueness identification
geo_address	STRING	address
geo_lng	FLOAT	longitude
geo_lat	FLOAT	latitude
geo_hash	STRING	geo hash, created by position
ipfs_hash	STRING	ipfs hash, created by saving data
addGeoInfoByParam()	FUNCTION	add position message
getGeoInfoByParam()	FUNCTION	get position message
mixGeoHashByParam()	FUNCTION	get GeoHash
addIpfsDataByParam()	FUNCTION	add ipfs data
mixIpfsHashByParam()	FUNCTION	associate ipfs data

1.2 Database Object Map

1.2.0 Chose Database

There is a test report of search 1 million poi data by some friends.

Database Name	Proc Time	Support Regional Search		Support Multi Condition
redis(3.2.8)	1-10ms	yes	no
mongo(3.4.4)	10-50ms	yes	yes
postgreSQL(9.6.2)	3-8ms	yes	yes
mysql（5.7.18）	8-15ms	yes	yes

compare these statistics data, I chose MySql to support the geo system demo.

• Older than MySql 5.7.4, the myISAM engine has Geom to implement this function.
• Newer than MySql 5.7.4, InnoDB engine also support for space index.

PS:

• 1. Not the best database, only have the suit database.
• 1. Researching the storage effiency, since the network node is test now and so instability that there have serious slow problem of transmission data, we can't reach high frequency to update and millions of data to index on IPFS.
• 1. Maybe we should find a efficiency plan to use IPFS under the current situation now.

1.2.1 Object Model Map To Table Struct

-- table struct `geo_object`
--

CREATE TABLE `geo_object` (
  `geo_id` bigint(20) NOT NULL  AUTO_INCREMENT,
  `geo_loc` point NOT NULL,
  `geo_address` varchar(255) NOT NULL,
  `ipfs_hash` varchar(255) NOT NULL
) ENGINE=MyISAM DEFAULT CHARSET=utf8 COMMENT='geo object model';

-- Indexes for table `geo_object`
ALTER TABLE `geo_object`
  ADD PRIMARY KEY (`geo_id`),
  ADD SPATIAL KEY `geo_loc` (`geo_loc`);

2. Demo

Base on these concept and design model above, now we can implement a simple demo:

2.0 Upload Position Data By IPFS

• The detail of deploy a single node of ipfs can refer to use ipfs to build a decentration wiki system

• Now the official support curl method, we can use addIpfsDataByParam() to implement RPC.

  curl -F file=@myGeoFile "http://localhost:5001/api/v0/add?recursive=false&quiet=false&hash=sha2-256"
  

• PS: This demo use location node to upload data, for protect the service available, we suggest use ipfs-cluster and the detail can refer to the article ipfs family II.

2.1 Get Data From IPFS and Relate it.

The response data is 'multipart/form-data' type. If success, these data would be recived:

{
    "Name":"myGeoFile"
    "Hash":"QmYftndCvcEiuSZRX7njywX2AGSeHY21Sa7VryCq1mK1Ew"
    "Bytes":"2428803"
    "Size": ""
}

After get the hash, we can use mixIpfsDataByParam() function to relate our Geo position data.

2.2 Run The Geo demo

Choice The First Base Geo Position (Simulate the position of user)

• latitude: 39.989049
• longitude: 116.313658

INSERT INTO `geo_object`(`geo_loc`, `geo_address`, `ipfs_hash`) VALUES (GeomFromText('POINT(39.989049 116.313658)'),'3W Coffee','QmYftndCvcEiuSZRX7njywX2A21Sa7VryCq1mK1Ew21')

Choice The Second Geo Position (Simulate the close position)

• latitude：39.988878

• longitude：116.313352

  INSERT INTO `geo_object`(`geo_loc`, `geo_address`, `ipfs_hash`) VALUES (GeomFromText('POINT(39.988878 116.313352)'),'Sourth Street of Zhong Guan Village','WCJIEFSCvcE231233HY21Sa7Vr1Cq1mK1Ew')

Choice The Third Geo Position(Simulate the far position)

• latitude：40.005466
• longitude：116.315938

INSERT INTO `geo_object`(`geo_loc`, `geo_address`, `ipfs_hash`) VALUES (GeomFromText('POINT(40.005466 116.315938)'),'Old Summer Palace','KBYftndCvcEiuSZRX7njyw1332Y21Sa723mKASDED')

Sphere Distance Algorithm

Assume the diagonal points of sphere fence is A1(x1, y1), B1(x2, y2):

x1 = lat + distance / ( 111.1 / COS(RADIANS(lng))),  
y1 = lng + distance / 111.1  

x1 = lat - distance / ( 111.1 / COS(RADIANS(lng))),  
y1 = lng - distance / 111.1  

// build the first order space filling curve
LineString(A1,B1)

PS:

• 1. One longitude of the equator is about equal 111.1km.
• 1. RADIANS() is a function to compute radians.
• 1. LineString() is a function to build the first order space filling curve.

2.3 Get The IPFS Data In A Region

Get The IPFS Data Within 1 KM Distance

SELECT  *  
    FROM    geo_object  
    WHERE   MBRContains  
                    (  
                    LineString  
                            (  
                            Point  
                                    (  
                                    39.989049 + 1 / ( 111.1 / COS(RADIANS(116.313658))),  
                                    116.313658 + 1 / 111.1  
                                    ),  
                            Point  
                                    (  
                                    39.989049 - 1 / ( 111.1 / COS(RADIANS(116.313658))),  
                                    116.313658 - 1 / 111.1  
                                    )   
                            ),  
                    geo_loc  
                    )

As shown in the following figure, we get the ipfs data within 1 km distance of 3W Coffee

Get The IPFS Data Within 10 KM Distance

SELECT  *  
    FROM    geo_object  
    WHERE   MBRContains  
                    (  
                    LineString  
                            (  
                            Point  
                                    (  
                                    39.989049 + 10 / ( 111.1 / COS(RADIANS(116.313658))),  
                                    116.313658 + 10 / 111.1  
                                    ),  
                            Point  
                                    (  
                                    39.989049 - 10 / ( 111.1 / COS(RADIANS(116.313658))),  
                                    116.313658 - 10 / 111.1  
                                    )   
                            ),  
                    geo_loc  
                    )

As shown in the following figure, we get the ipfs data within 10 km distance of 3W Coffee

PS: About this demo detail, we will write a new article to describe these.

3 Application Scene

• Vevue: A dapp to encourage user to take a picture on special location and reward some token.
• sign in by position: only arrive special location can reward some things or get some proof.
• money in store: attract people to go to some store and reward some money.
• protect parking space: protect some parking space that someone has bought.
• unite with AR game: AR game unite with position and reware some token.
• IOT: some IOT have position property, these may unite with this system.

4 Open Source Plan

goal: Expect everyone can realize some application scene by ipfs even block chain. Use technology to crate real value. Welcome everyone to join.

IPFS-Geo

meaning: This is a intelligent IPFS object and it's meta data has Geo property and support fast index 10 millions space data. It also provides LBS function api.

reference

• Wikipedia GeoHash
• block chain blog by YueChunLi
• GeoHash core theory
• 【application】use ipfs to build a decentration wiki system
• Vevue：a DApp to take photos in special position and reward tokens